Speaker: Joseph Buonomo, University of Texas at Arlington
Topic: Scalable Native Peptide Sequencing Via Innovation in “Click” Chemistries and Large Language Models
Date: May 12, 2025
Time: 6:15 pm Dinner, 7:15 pm Presentation
Location: Shimadzu Scientific Instrument, Inc. Training Center 7100 Riverwood Drive, Columbia, MD 21046 (Directions)
Dinner: Please RSVP to Dingyin Tao (owendtao@gmail.com) by Friday, May 9 if you will be attending the dinner.
Abstract: We are developing a groundbreaking method to sequence native peptides, especially for liquid biopsies, a less invasive way to detect cancer from bodily fluids like urine and blood. By accurately sequencing proteins from these samples, we can improve early cancer detection, monitoring, and personalized treatment, offering a powerful tool for doctors and researchers alike. Our approach uses a sophisticated sequencing platform combined with advanced artificial intelligence (AI). Imagine the process like a game of hangman, where you guess a word by uncovering one letter at a time. Here, we identify specific amino acids—akin to letters—in a protein. Instead of revealing all twenty amino acids at once, we focus on identifying eight key ones: K, Y, C, M, W, R, D, and E with highly chemoselective reactions. These amino acids are like the revealed letters in hangman. With the help of AI, acting as a super-smart guesser, we can infer the remaining twelve amino acids. The AI, trained on vast amounts of protein data, uses patterns and context to predict the full sequence, similar to how a hangman player might guess the entire word after seeing a few letters. Our method utilizes advances in bioconjugation and “click” chemistries, as well as unique barcoding strategies to encode cycle numbers and amino acid identities to accurately identify the eight key amino acids and their modifications. By leveraging these innovations, we can extend the read lengths of sequences and improve accuracy. We have designed our platform to be agnostic of detection method, applicable to DNA sequencing, nanopore, and fluorescent detection workflows.
To fully harness the potential of AI beyond sequencing, we are also exploring its application in selecting generic drug suppliers for clinical and research purposes. Reliable sourcing is a persistent challenge, especially as the pharmaceutical supply chain grows more complex and globalized. Traditional vetting processes often rely on manual document reviews, regulatory audits, and personal networks, which are time-consuming and susceptible to human error. Our approach involves training AI algorithms on comprehensive datasets including supplier certifications, manufacturing history, recall records, and even real-time logistics data. This allows us to systematically evaluate suppliers, flag inconsistencies, and predict future reliability based on historical patterns. Integrating AI-driven supplier selection into our sequencing workflow ensures not only higher confidence in reagent quality but also scalability and traceability in our operations. This layer of intelligence helps safeguard the reproducibility of our results and ultimately supports more robust diagnostic platforms.
